Research Articles

Writing this reply presents a challenge for us since the authors of the preceding article reinforce most of our substantive points published previously (Brace et al. 2004, 2006). In this new article, Berry et al. (2007) acknowledge that there is more cross-sectional than longitudinal variation in the Berry et al. (1998) measure of state citizen ideology and imply that researchers should take precautions not to confuse them. They admit that their measure is tapping something different from survey-based self-identification, and they also note the distinction between absolute and relative change in their ideology measure. Given our many points of agreement, this reply focuses on four key points made by Berry et al. in the preceding article on which we differ. First, we address the argument that focusing on intra-state attitude change ignores important attitude change affecting all states simultaneously. Second, we discuss Berry et al.'s distinction between symbolic ideology and operational ideology. Third, we consider the validity and reliability of the various measures. Finally, we revisit our choice of statistical methods. Although we certainly do not conclude that Berry et al.'s measure lacks clear utility or relevance, we believe it may serve a more limited purpose than the authors intended. We urge researchers to consider the conceptual, methodological, and epistemological concerns raised in our response before using any indicator of state public opinion.